Thanks for being a member!
Below, I’m posting all my preseason football newsletters that will go out to the free email list. I’m including them here so I don’t have to send you those annoying “become a member” pitches like I do on Saturday.
Buffalo Bills
Sports books win because of the hold. To win $100, you have to bet $110 on a typical spread or total.
Wouldn’t life be much better if you could bet $100 to win $100? The sports book would no longer have an edge.
In the book The Logic of Sports Betting, Ed Miller and Matthew Davidow discuss these no hold markets. While no single book will let you have +100 odds on both sides of a spread, it is possible to find the equivalent or better across two sports books.
As an example, Circa has a regular season win total of 10.5 for the Buffalo Bills. The price to bet the over is -150, so you bet $150 to win $100.
At The Power Rank, I use the NFL win totals market at Circa and back out a rating for each team. For Buffalo, this considers their difficult division and a schedule that also has Philadelphia, Kansas City and Dallas.
The Unabated NFL simulator provides a unique opportunity to compare the market at Circa to other sports books. Based on my preseason market ratings, this tool simulates the season 10,000 times to generate probabilities for a multitude of events.
For example, the win total market at Circa translates to a 51.5% chance that Buffalo wins the AFC East. Unabated shows this price of +106 against the posted odds at a number of sports books.
DraftKings has Buffalo at +120 to win the division. Between Circa and DraftKings, you have a negative hold market. This gives you two choices:
- If you like Buffalo, bet +120 at DraftKings to win the division.
- If you think Buffalo is overrated, bet under 10.5 wins, +130 at Circa.
In The Logic of Sports Betting, the authors recommend taking a side in a no hold market. I am high on Buffalo for three reasons.
First, my market rankings most likely underestimate the probability for the Bills to win the division. In an earlier episode of the 2023 preview series, I discussed how the New York Jets are overrated in the market. In addition, Miami QB Tua Tagovailoa has yet to play an entire NFL season due to injury.
Second, Buffalo has been excellent the past two seasons. The Bills have had the best point differential per game during this time. QB Josh Allen and head coach Sean McDermott have played a pivotal role in both seasons.
The excellence of the Bills might seem diminished with a bad 27-10 loss to Cincinnati in the playoffs last season. However, don’t let one game fool you.
Third, which sports book do you trust more? A sharp book like Circa takes large bets from anyone, even professional bettors. A public book like DraftKings seems to specialize in restricting bets from anyone who wins.
There is value in Buffalo +120 to win the AFC
Ohio State
Ohio State enters the season at #3 in the preseason AP poll. This might seem like a stretch after what happened against Michigan last year.
In Columbus, Michigan scored a convincing 45-23 road win. Even worse than the scoreboard was the way it happened.
With about seven minutes remaining in the game, Michigan had a 31-23 lead. This was a typical situation in which the team with the leads wants to take time off the clock.
Ohio State knew this, and they had seven defenders near the line of scrimmage. These conservative run plays usually go the way of the defense. Except it didn’t this time.
Michigan RB Donovan Edwards found a seam, and the single high safety took a bad angle. Edwards ran around him for a 75 yard touchdown.
A few minutes later, Michigan had the ball again with a 3rd and 3. The offense and defense were in the same formation as the last touchdown with a single difference: Ohio State brought the high safety near the line of scrimmage.
The formation change didn’t matter. Edwards found a hole and scored an 85 yard touchdown run.
These plays reflected poorly on Ohio State’s defense. However, there is a huge random element in explosive plays. The research goes back to seminal work by ESPN’s Bill Connelly.
For explosiveness, Bill looked at what amounts to yards per play on successful plays (A play is a success based on the definition used in looking at Aaron Rodgers and the New York Jets). In college football, he found zero correlation between different parts of the season. I’ve also found very little predictability in explosiveness in the NFL.
The talking heads on broadcasts will rant about how the defensive scheme led to those big Michigan runs. They might lament the aggressiveness of Ohio State DC Jim Knowles, ignoring that his defense was mostly great. They ranked 4th in my adjusted success rate in 2022.
While scheme matters, the talking heads will not talk about the random element in explosive plays. A few plays against Michigan mask the massive improvement the defense made in the first year under Knowles. They ranked 45th in adjusted success rate in 2021 before he arrived.
Ohio State’s struggles with explosive plays are clear when comparing their rank on success rate (4th) versus yards per play (49th). However, Knowles did not have the same discrepancy in 2021 at Oklahoma State (5th in success rate, 4th in yards per play). With the talent that Ohio State returns, the defense should continue to improve.
On offense, Ohio State must replace QB CJ Stroud, the 2nd overall pick of the NFL draft. Kyle McCord will start in the first game against Indiana, but he has not won the job over Devin Brown.
However, coach Ryan Day has an excellent track record on offense. Ohio State has an elite set of receivers, headlined by Marvin Harrison Jr. These receivers got so much separation against Michigan State last year they could have used a catapult to deliver the football.
A variety of trusted preseason metrics like the Buckeyes:
- 3rd in my market rankings based on win totals
- 2nd in SP+, a metrics based approach by Bill Connelly
Don’t let a few plays against Michigan fool you. Ohio State is not overrated at #3 in the preseason AP poll (I’ll give you my third overrated team on Saturday).
Michigan
It is easy for Michigan fans to dream this season. QB JJ McCarthy returns for his second year as a starter, and he has the talent to win the Heisman.
Even if McCarthy doesn’t reach this ceiling, consider how Michigan’s QB situation compares with other top contenders.
- Georgia will start redshirt sophomore Carson Beck, who has 58 career pass attempts. They also replaced OC Todd Monken with former OC Mike Bobo.
- Ohio State will start Kyle McCord against Indiana on Saturday, but he has not won the job over Devin Brown.
- Alabama is considering Jalen Milroe, Ty Simpson and Notre Dame transfer Tyler Buchner at QB heading into their first game.
In addition, Alabama and Ohio State lost their QB to the first two overall picks of the NFL draft. Georgia will no longer have Stetson Bennett, the most accomplished QB in program history with back to back championships.
Right now, Michigan has a better QB situation than these teams. In 2023, this is the true source of dreams for Michigan fans, a group that I’m a part of here in Ann Arbor.
Then, the preseason AP poll put Michigan at #2. That is most likely too high, as Michigan does have concerns.
The offense should be great to exceptional. Michigan will be able to bludgeon inferior opponents with the run game. They should have an exceptional offensive line and one of the best RB tandems in college football with Blake Corum and Donovan Edwards.
The question is how much the offense lets McCarthy throw to receivers, a relative weakness on this team. WRs Cornelius Johnson and Roman Wilson return but need to improve on their total of 57 receptions from 2022.
The defense was good enough last season, but they ranked only 32nd in my success rate adjusted for opponent. Now, they must replace the leading pass rusher and two cornerbacks.
CB Will Johnson will be fantastic, but Michigan has questions at the opposite CB position. Edge Jaylen Harrell returns after having the second highest tackle for loss and sack totals last year, but the defense will need another pass rusher to emerge.
For the second year in a row, Michigan faces an out of conference schedule softer than a baby’s bottom. The early conference games are only marginally more challenging. Michigan should be 9-0 heading into a road trip to Penn State. They end with Ohio State at home in Ann Arbor.
To justify 2nd in the preseason poll, Michigan must make the championship game. The market at Caesars implies a 44% chance to make the playoff.
Let’s assume that Michigan has a 50-50 chance to win the playoff semi-final. That leaves roughly a 1 in 5 chance that Michigan lives up to #2.
Georgia, Ohio State and Alabama all have higher national championship odds because of program history. For example, consider what Bob Stoll said on the last episode of The Football Analytics Show.
By his compensated yards per play, Georgia has had a top 10 offense in five of the past six seasons. They have started three different QBs during this tenure. He doesn’t expect that to change with Carson Beck.
While Michigan should definitely be a top 5 team, they are overrated at #2 in the preseason poll.
TCU
Last season, Sonny Dykes took over a TCU program that had parted with longtime legendary coach Gary Patterson. With the uncertainty, the Horned Frogs were picked 7th of 10 teams in the preseason Big 12 poll.
Despite these low expectations, Dykes took TCU to the college football championship game against Georgia. While they got routed in that game, the program vastly exceeded expectations in 2022.
TCU comes into 2023 with much higher expectations. However, let’s not forget the good fortune they had a year ago. Let’s count down the top 5.
- 5. Down two against Baylor, TCU ran the field goal unit onto the field and kicked a game winner as time expired. Dykes talked about how they practice this play, but there was no room for error as the previous play left :17 on a running clock with no timeouts.
- 4. Michigan gifted TCU two interceptions returned for a touchdown in a playoff semi-final. TCU won by six points despite having fewer yards than Michigan.
- 3. Kansas QB Jalon Daniels, a Heisman contender at the time, left the game against TCU with an injury and didn’t return. TCU came back to win 38-31.
- 2. TCU had a 5-0 record in one score games during the regular season. Record in one score games tends to regress to .500, and regression will have a much different effect on TCU than Oklahoma (0-5 in 2022).
- 1. The backup quarterback almost wins the Heisman.
That’s right, Max Duggan was not the starter when TCU took the field against Colorado to open the season. Chandler Morris won the job but got hurt in the first game.
Duggan had an incredible season and finished second in the Heisman voting. Now, the Los Angeles Chargers have waived him and only kept two QBs.
Heading into 2023, TCU has Morris to replace Duggan. This is a piece of good news amidst plenty of roster turnover.
Morris will not have the services of TCU’s top three receivers from last season. This includes NFL first round draft pick Quentin Johnston.
The news might be worse on defense. They lost their most important player on all three levels of the defense.
Sonny Dykes has excelled at using the transfer portal to upgrade his roster, both last year at TCU and the previous four years at SMU. He will need help again in 2023, as players like Oklahoma State WR JP Richardson (49 receptions, 503 yards in 2022) must make immediate contributions.
Even if they do get the transfer bump, TCU is unlikely to have the same good fortune as last season. They are overrated 17th in the preseason AP poll. My market rankings, which take season win totals and back out a rank and rating for each team, has TCU at 26th.
This overrated prediction might look silly in the early part of the season, as TCU should start 5-0 against weaker competition. They should be favored in the next two games at Iowa State and against BYU.
However, my numbers make them the underdog in four of their last five games. Don’t be surprised if they fall out of the top 25 during this difficult stretch.
How to predict NFL interceptions
In the 2nd quarter of the Super Bowl, Philadelphia QB Jalen Hurts throws the ball deep down the left sideline. In a tie game, Hurts was looking for a big gain to AJ Brown against Kansas City.
However, Brown was double covered, and Kansas City’s Juan Thornhill got his hands on the ball. In one possible world, the ball drops into the hands of Trent McDuffie, the other defender in coverage. Maybe he finds a seam down the sideline and scores a touchdown.
Instead, the ball fell harmlessly to the turf. While interceptions can have a massive impact on a football game, this play shows the random element in picks.
To put some numbers behind this, consider a quarterback’s interception rate, or interceptions divided by pass attempts. An NFL quarterback’s interception rate one season has little ability to predict interception rate the following season (R-squared value of 7%).
However, you can make better predictions for interceptions by expanding the set of events. As suggested earlier, the key idea is looking at plays in which a defender gets a hand on a pass.
The NFL play by play data tracks passes defended, or incomplete passes in which a defender gets a hand on the ball or jars it loose with a hit. To predict interceptions, consider bad balls, or the sum of interceptions and passes defended. These are the plays in which the QB put the ball in a dangerous position.
Bad ball rate is bad balls divided by pass attempts. From one season to the next, bad ball rate has an R-squared of 27%, about as predictive as QB metrics get.
In addition, the fraction of bad balls that turn into interceptions is random from one season to the next. There is strong regression to the NFL average of 19.7%.
If you are going to remember one thing, it should be this: quarterbacks have control over putting the football into dangerous situations. However, once they put the ball in danger, they have no control over whether the play ends as an interception.
To see the implications for the 2023 season, let’s look at New York Giants QB Daniel Jones and compare his rates with NFL averages (interception rate of 2.3%, bad ball rate of 11.6% over the past three seasons).
- 2020: interception 2.2%, bad ball 13.4%.
- 2021: interception 1.9%, bad ball 10.5%.
- 2022: interception 1.1%, bad ball 10.7%.
First, note that Jones did not excel at interception prevention because new head coach Brian Daboll showed up in 2022. Despite an injury riddled season in 2021, Jones had a better than NFL average bad ball rate.
Second, Jones will not be able to sustain a remarkable 2022 in which he only threw six interceptions. I’ve found that a three year average of bad ball rate is optimal for predicting the next season.
Over the past three seasons, Jones has had a 11.6% bad ball rate, right at NFL average. He will not be able to sustain his excellent turnover prevention in 2023.
To put this in a different way, Jones had 11.7% of his bad balls in 2022 end up as interceptions. This rate will regress to the NFL average of 19.7% in 2023.
My calculations based on bad ball rate give 10.8 interceptions for Daniel Jones this season. Although an injury can lower this estimate, it suggests value in over 8.5 interceptions (-110 available at DraftKings).
To check out bad ball data for all NFL QBs in 2022, click here.
New York Jets
The New York Jets are riding the hype train heading into the 2023 season.
The Jets acquired QB Aaron Rodgers to complement a defense that killed it last season. With a win total of 9.5, the market expects the Jets to make the playoffs and possibly push Buffalo for the AFC East division title.
However, there are some warning signs about this team.
First, there are questions about Aaron Rodgers. He will turn 40 years old this season, and his play fell off in Green Bay last season.
In my research, I’ve found success rate on passing plays as the most predictive metric for an offense. While there are multiple definitions of success rate, I use a simple criteria based on the yards the offense gains towards the next first down.
A play is a success if the offense gains the following fraction of these necessary yards:
- 50% on 1st down
- 70% on 2nd down
- 100% on 3rd, 4th down
The Power Rank specializes in taking this success rate for a QB and adjusting for opposing defenses. Based on this calculation, Rodgers with the Packers last season was predicted to have a passing success rate of 42.3% against an average NFL defense. This is below the NFL average of 42.8%.
Rodgers might have struggled because of a thumb injury that never seemed to heal last season. In addition, Green Bay had injuries at wide receiver.
With the Jets, Aaron Rogers will now have the services of WR Garrett Wilson. The second-year player out of Ohio State had 1.85 yards per route run last season compared to an NFL average of 1.5 for wide receivers.
In addition, Wilson put up these numbers with the QB incompetence of Zach Wilson, Joe Flacco and Mike White.
After Wilson, Rodgers will throw to his old buddies from Green Bay. Alan Lazard and Randall Cobb both rate roughly NFL average by yards per route run.
The Jets offense projects as NFL average. While this might seem like a knock on Aaron Rodgers, it’s actually a massive improvement from last season.
On defense, the Jets excelled in 2022. In my passing success rate adjusted for opposing QBs, the Jets ranked 3rd.
A huge reason was cornerback Sauce Gardner. The rookie out of Cincinnati had a PFF coverage grade of 90, an elite number on a scale of zero to a 100.
However, the defense had the good fortune of having the lowest negative impact due to injuries. This comes from a Football Outsiders metric that gives more weight to starters than backups.
In addition, the top three cornerbacks for the Jets did not miss a single game. This includes DJ Reed and Michael Carter in addition to Sauce Gardner. With the randomness inherent in injuries, the Jets are unlikely to have the same health and hence performance in 2023.
In addition, the Jets have a difficult schedule. In addition to two games against Buffalo and Miami, they also play the following teams:
- At Dallas, Week 2
- Kansas City, Week 4
- Philadelphia, Week 6
- Los Angeles Chargers, Week 9
I like the Jets under 9.5 wins (+110 at most sports books).